A review of Euryoryzomys legatus (Rodentia, Sigmodontinae): morphological redescription, cytogenetics, and molecular phylogeny

The taxonomic history of Euryoryzomys legatus has been complex and controversial, being either included in the synonymy of other oryzomyine species or considered as a valid species, as in the most recent review of the genus. Previous phylogenetic analyses segregated E. legatus from E. russatus, its putative senior synonym, but recovered it nested within E. nitidus. A general lack of authoritative evaluation of morphological attributes, details of the chromosome complement, or other data types has hampered the ability to choose among alternative taxonomic hypotheses, and thus reach a general consensus for the status of the taxon. Herein we revisit the status of E. legatus using an integrated approach that includes: (1) a morphological review, especially centered on specimens from northwestern Argentina not examined previously, (2) comparative cytogenetics, and (3) phylogenetic reconstruction, using mitochondrial genes. Euryoryzomys legatus is morphologically and phylogenetically distinct from all other species-level taxa in the genus, but its 2n=80, FN=86 karyotype is shared with E. emmonsae, E. nitidus, and E. russatus. Several morphological and morphometric characters distinguish E. legatus from other species of Euryoryzomys, and we provide an amended diagnosis for the species. Morphological characters useful in distinguishing E. legatus from E. nitidus, its sister taxon following molecular analyses, include: larger overall size, dorsal fur with a strong yellowish brown to orange brown tinge, flanks and cheeks with an orange lateral line, ventral color grayish-white with pure white hairs present only on the chin, presence of a thin blackish eye-ring, tail bicolored, presence of an alisphenoid strut and a well-developed temporal and lambdoid crests in the skull, and a labial cingulum on M3. Molecular phylogenetic analyses recovered E. legatus as a monophyletic group with high support nested within a paraphyletic E. nitidus; genetic distances segregated members of both species, except for an exemplar of E. nitidus. Our integrated analyses reinforce E. legatus as a full species, but highlight that E. macconnelli, E. emmonsae, and E. nitidus each may be a species complex and worthy of systematic attention. Finally, we also evaluated the chromosome evolution of the genus within a phylogenetic context.

One species of Euryoryzomys with a complex taxonomic history is E. legatus. This taxon is distributed across premontane and montane forests along the eastern Andean slopes in central and southern Bolivia (Chuquisaca, Santa Cruz, and Tarija departments; the type locality is Caraparí, Tarija, Bolivia; Thomas, 1925) and northwestern Argentina (Jujuy and Salta provinces), at elevations ranging from 500 to 2,100 m (Anderson, 1997;Teta et al., 2007;Percequillo, 2015;Pardiñas & Ruelas, 2017). Sympatry with E. nitidus is observed in eastern Andean areas of central Bolivia (Percequillo, 2015). Since the original description by Thomas (1925) as Oryzomys legatus, the taxon was considered as a valid species by several authors (e.g., Gyldenstolpe, 1932;Tate, 1932;Ellerman, 1941) until, without any supporting evidence, Hershkovitz (1960) placed it as a synonym of Oryzomys laticeps, and Cabrera (1961) listed it as a subspecies of O. capito-both capito and laticeps are currently considered as junior synonyms of the oryzomyine Hylaeamys megacephalus; see Brennand, Langguth & Percequillo (2013). Massoia (1975) again considered O. legatus as a valid species but, shortly thereafter, Gardner & Patton (1976), followed by Honacki, Kinman & Koeppl (1982) and Redford & Eisenberg (1992), viewed this nominal form as a junior synonym of O. nitidus. Once again, the validity of O. legatus was sustained by Musser & Carleton (1993), but Musser et al. (1998) later subsumed it in the synonymy of

Cytogenetic analyses
We obtained chromosomal preparations from one female (CML13250) and one male (CML13251) of E. legatus. Preparations were obtained in vivo from bone marrow and spleen, following the protocols of Ford &Hamerton (1956) andYonenaga (1972), with modifications. Slides were Giemsa stained, and CBG and GTG-banding patterns were obtained according to Sumner (1972) and Seabright (1971), respectively, after modifications. Fluorescence in situ hybridization (FISH) with telomeric probes labeled with FITC was carried out following the recommended protocol (Telomere PNA FISH Kit/FITC, Code No. K5326, DAKO). Slides were counterstained with 4 ,6-Diamidino-2phenylindole dihydrochloride (DAPI) in antifade mounting medium (Vectashield with DAPI, Vector).
We analyzed 72 metaphases from the male and 56 from the female specimen to establish both diploid (2n) and fundamental numbers (FN = number of arms of the autosomes). Metaphases were digitally captured in an Axioskop 40 epifluorescence microscope (Carl Zeiss) equipped with an Axiocam camera and AxionVision software. Adobe Photoshop CS5.1 was used for assembling the karyotypes.

Phylogenetic and genetic distance analyses
Molecular data consisted of partial sequences from two mitochondrial genes, 801 bp of the cytochrome-b (cyt-b) and 667 bp of the cytochrome oxidase I (coxI ; Table S1). In addition to newly generated sequences for 40 specimens, we also acquired sequences from seven individuals from GenBank (Table S1). DNA was extracted from liver or muscle using Chelex 5% (Walsh, Metzger & Higuchi, 1991). Cytochrome-b was amplified with primers MVZ05 and MVZ16 (Smith & Patton, 1993), and cytochrome oxidase I was amplified with primers LCO1490 and HCO2198 (Folmer et al., 1994). Polymerase chain reactions of 15 µL or 25 µL consisted of 30 ng of DNA, 10 mM of each primer, 0.2 mM of dNTP, reaction buffer (50 mM KCl, 2.5 mM MgCl2 and 10 mM Tris-HCl; ph 8.8) and 0.2 units of Platinum Taq DNA polymerase (Invitrogen). Amplification cycles were performed with an initial denaturation at 94 • C for 5 min; 30 cycles for cyt-b and 35 for coxI composed of: denaturation at 94 • C for 30 s, 45 s of annealing at 48 • C, and extension at 72 • C for 45 s; and a final cycle of extension at 72 • C for 5 min. Sequencing was performed at Laboratório de Bacteriologia II, Instituto Butantan, São Paulo, Brazil. We were unable to obtain sequences for both markers for some individuals (Table S1). The sequences were edited using Geneious R7 (Kearse et al., 2012) and deposited in GenBank (Table S1).

Morphometric analyses
The following standard external measurements were recorded from field catalogs and tags: TL: total length; T: tail length; HF: hind foot length (including claw); E: ear length; and W: body mass. The following skull measurements were recorded with a vernier caliper to the nearest of 0.01 mm, following Hershkovitz (1962), Myers (1989) and Myers, Patton & Smith (1990): CIL: condyloincisive length; PB: palatal bridge; RL: rostral length; OL: orbital length; RW2: mid rostral width; ZP: zygomatic plate depth; IOC: interorbital constriction; ZB: zygomatic breadth; BB: braincase breadth; OCW: occipital condyle width; DL: diastema length; MTRL: maxillary toothrow length; IFL: incisive foramina length; AW1: alveolar width (across external side of both M1); BLLT: bullar length less tube; ML: mandible length. Five age classes were defined according to tooth wear following the descriptions in Jayat et al. (2020) and Fig. S1. Descriptive morphometric and univariate comparisons (Tukey's pairwise comparison; Table 1) for samples of the species of Euryoryzomys were carried out with the software PAST (Hammer, Harper & Ryan, 2001). Specimens of age classes 2 and 3 (the largest pool of available specimens for the whole specimen sample) were pooled in tests of significant size differences (for both, P ≤ 0.05 and P ≤ 0.01).
With the aim of reducing the dimensionality of morphometric data in comparing E. legatus with other 5 species of Euryoryzomys, we conducted two Principal Components Analyses (PCA) using only skull measurements. The first PCA explored skull size variation (''size PCA'' hereafter; Table 2) and we used only specimens of the age classes 2 and 3 (again taking advantage of the largest pool of available specimens). The second PCA was developed to explore skull shape differences (''size free PCA''; Table 3) and all age classes were used (after removing the size effect). This latter analysis was conducted using Mosimann shape variables (Mosimann & James, 1979) obtained as described in Meachen-Samuels and Van Valkenburgh (2009). Principal components (PC) and their statistical significance were obtained as in Jayat et al. (2020). Finally, we conducted a Discriminant Function Analysis using Mosimann shape variables and specimens of all age classes (''size free DFA''; Table 4). All multivariate statistical analyses were conducted in PAST (Hammer, Harper & Ryan, 2001) and using only the specimens without missing data.

Morphological description
We re-described the skin color pattern and the skull of E. legatus and compared them with parapatric populations assigned to E. nitidus (its putative sister species) and other species of the genus. Terminology used to describe skull anatomical features followed   Hill (1935), Carleton (1980), Voss (1988), Carleton & Musser (1989), Steppan (1995), andWeksler (2006). Descriptions of the molar cusp pattern followed Reig (1977).

Cytogenetic analyses
Both karyotyped individuals presented 2n=80 and FN=86. The karyotype is composed of 35 acrocentric pairs, with pair 1 the largest of the complement and others pairs (2 to 35) decreasing gradually in size, and four small metacentric pairs (36 to 39). The sex chromosomes are readily distinguishable from the autosomes because of their different morphologies: the X is a large submetacentric and the Y is a medium-sized submetacentric ( Fig. 2A). CBG-bands revealed subtle pericentromeric constitutive heterochromatin in all autosomes. The X-chromosome is heterochromatic in the short arm, and the Y is entirely heterochromatic (Fig. 2B). GBG-bands allowed the identification of the homologues (Fig. 3). FISH showed signals exclusively at the telomeric regions of all chromosomes (Fig. 4A) and DAPI evinces the Y strongly and entirely stained (Fig. 4B).

Phylogenetic analyses and genetic distances
The phylogenetic analyses of cyt-b provided high nodal support for the monophyly of Euryoryzomys (Clade A, BI: 0.93, ML: 95%) (Fig. 5  The phylogenetic analyses of the concatenated cyt-b and coxI markers (Fig. 6) showed Euryoryzomys as a monophyletic group (Clade A) with high support (BI: 0.99, ML: 99%),

Morphological comparison of E. legatus
Externally, E. legatus specimens show all the character states reported by Thomas (1925) in the original description for the species. These specimens are similar to those of other species of Euryoryzomys in having a strong countershading between the dorsal and ventral pelage coloration. Nevertheless, individuals of E. legatus generally have dorsal fur with a strong yellowish brown to orange brown tinge, with clearer and brighter flanks and cheeks, forming an orange lateral line in most of the specimens examined; and ventral color grayish-white, with pure white hairs present only in small area on the chin. This more intense ''ochraceous highlights'' of the pelage, mentioned by Musser et al. (1998) as a difference between specimens of E. legatus and those of E. nitidus, may be also a useful character in separating specimens of E. legatus from those of E. lamia, which have a lighter dorsal coloration. This latter species also differs from E. legatus in ventral coloration, being more grayish cream than whitish. Individuals of E. nitidus possess unicolored tails (Musser et al., 1998;Percequillo, 2015), while in almost all the specimens of E. legatus the tail is certainly bicolored (dark brown above and grayish-white below). The presence of a thin blackish eye-ring in E. legatus was not described for other species of Euryoryzomys, so this character may be also useful in separating this form.
The skull of specimens of E. legatus follows the general form for the genus, with a long rostrum, narrow interorbital region with lateral margins divergent posteriorly, and the zygomatic arches widest at their squamosal roots. The mandible is deep and robust, with coronoid and condylar processes somewhat equal in height (separated by a shallow superior notch), and the angular process not extending posteriorly behind the condyloid process, also resembling the condition in other species of the genus. Qualitative characters of the skulls of E. legatus and other species the genus are very similar (see Table 13 in Weksler, 1996). Nevertheless, an alisphenoid strut, present in all specimens of E. legatus (including the holotype), is found in about half of previously examined specimens of E. nitidus (see Musser et al., 1998), and it is present in most, but not all, specimens of E. lamia and E. russatus, and completely absent in specimens of E. macconnelli. Additionally, we observed well-developed temporal and lambdoid crests in E. legatus, which were described as not well developed in E. nitidus (Weksler, 1996). Specimens of E. legatus with unworn molars always show a labial cingulum on M3, which is well developed in most of the examined specimens; in contrast, a labial cingulum is absent or vestigial in E. lamia and E. nitidus (Weksler, 1996).

Morphometric analyses
Descriptive statistics for each of the species of Euryoryzomys are summarized in Table 1. Euryoryzomys legatus differed in several of the recorded metric characters from all other species of the genus. Euryoryzomys lamia, the most similar to E. legatus following the univariate comparison, showed significant differences in seven of the 21 morphometric characters studied. On the other extreme, E. nitidus was the most different species when compared to E. legatus (these species significatively differed in 18 of the 21 analysed measurements).
The first 3 principal components of the ''size'' PCA (  (Fig. 7A). Specimens of E. emmonsae occupy negative values on PC I (being characterized by narrower zygomatic plates and smaller skulls) and those of E. lamia occupy positive values (broader zygomatic plates and larger skulls). The other four species widely spread over the PC I, but E. macconnelli and E. nitidus were mostly on the negative side. PC II mostly separated specimens of E. macconnelli from the other species, mostly by metric characters as the palatal bridge (longer in E. macconnelli) and the incisive foramina length (shorter in E. macconnelli).
The ''size free'' PCA (Table 3) shows a similar general pattern to that observed in the ''size'' PCA, but E. legatus better separated from E. lamia and E. macconnelli (Fig. 8B). The first 3 principal components (all judged statistically significant according to the Broken Stick test) summarized 68.3% (PC I 34.3%, PC II 19.3%, and PC III 14.8%) of the explained variance. E. lamia occupied negative values on PC I (Fig. 7B), being the most specimens characterized by zygomatic plates (ZP) relatively broad and short molar series (MTRL). E. legatus mostly occupied the positive side of this PC (comparatively small ZP and large MTRL). PC II once again separated specimens of E. macconnelli from other species of Euryoryzomys, mostly by the palatal bridge (PB) and the incisive foramina length (IFL).  Table 3.
(C) Individual specimen scores based on log-transformed values of 16 cranial measurements (Mosimann shape variables) projected onto the first and second discriminant functions of the ''size-free'' DFA. Character loadings and the variance explained by each of the first four discriminant functions appear in Ta (Fig. 7C). DF I mostly segregated E. macconnelli from all other species on the basis of palatal bridge (PB) and diastema length (DL). DF II separated E. legatus from other species of Euryoryzomys except E. russatus. Incisive foramina length (IFL), interorbital constriction (IOC), and braincase breadth (BB) were important metric characters in this separation. The percentage of correct classifications following the jackknifed confusion matrix in this analysis was high (Table 5). Only 9.6% of the specimens of E. legatus were misclassified.
Given all the evidence, we consider E. legatus as a valid taxon and provide a taxonomic account and an emended diagnosis.
Morphological description: External measurements for adults (age classes 4 and 5) are: total length 281-340 mm; tail 132-161 mm; hindfoot 31-37 mm; ears 23-26 mm; body weight 57-102 g (Table 1). Skin with dorsum yellowish brown to orange brown (Fig. S2); individual hairs plumbeous at base and orange at tip (underfur), or plumbeous at base and black at the tip (guard hairs). Flanks and cheeks clearer than dorsum, bright orange in most of the specimens examined. Belly strongly contrasting with the rest of the body, predominantly whitish, with hairs gray based and tipped whitish; most examined specimens possess a small white spot (all white hairs) on the chin. Eyes surrounded by a thin blackish ring. Ears darker than dorsum, internally and externally covered with brown hairs. Tail bicolored, dark brown above and grayish-white below, with scales evident without magnification. Hand and feet covered dorsally by whitish hairs, with a tuft of white hairs over the claws.
Skull elongated and slightly compressed laterally (Figs. S3 and S4). Rostrum long and robust, nasals extending anteriorly well ahead of the anterior face of upper incisors and the gnathic process, and posteriorly not extending beyond the level of the lacrimals (Figs. 8A and 8B). Nasolacrimal capsules well developed, except in very young specimens. Zygomatic notches excavated, usually as wide as deep, but deeper than wider in some individuals. Lacrimals well visible, generally large and pointy laterally, except in very young and some adult specimens examined (Fig. 8A). Interorbital region posteriorly divergent, with dorsolateral margins beaded showing an overhanging shelf in most specimens examined (even vertically raised in some individuals; Fig. 8C). Frontoparietal suture predominantly U-shaped (Fig. 8C). Interparietal large, approximately 2.5 to 3 times wider than long (Fig. 8D). Zygomatic arches slightly convergent anteriorly and not well expanded laterally. Zygomatic plates comparatively broad and with straight or slightly concave anterior margins (Figs. 8B and 8E). Braincase relatively small, with temporal and lambdoidal crests well visible in most adult specimens (age classes 3 to 5; Fig. 8D). Interpremaxillary foramen is small and rounded (Fig. 8E). Incisive foramina proportionally short, posteriorly not extending beyond the anterior face of the procingulum of M1 (Fig. 8E). Bony palate extending behind the level of the alveolus of M3, with large posteropalatal pits -located slightly ahead, or at the same line, with the anterior margin of the mesopterygoid fossa -and with small palatine excrescences (Fig. 8F). Mesopterygoid fossa with the roof completely ossified, approximately of the same width of the parapterygoid fossae, with a rounded anterior margin, and without a well-developed median spine (Fig. 8F), except for a few specimens examined which shows a small median spine. Parapterygoid fossae somewhat excavated, with straight or divergent backwards lateral margins (Fig. 8F). Tympanic bullae small and with large and flattened Eustachian tubes (Fig. 8G). Alisphenoid strut present in all the specimens examined. Hamular process of the squamosal large, slender in most of the specimens, and distally pointy (Fig. 8H). Postglenoid foramen larger than subsquamosal foramen (Fig. 8H). Carotid arterial supply follows the Patterns 1 of Voss (1988), with the common carotid artery bifurcated behind the auditory bulla to form the external and internal carotid arteries, and the stapedial artery split into infraorbital and supraorbital branches.
Coronoid process of the mandible approximately at the same level as the condyloid process (Fig. 8I), but slightly above or below in some specimens examined. Angular process generally do not extend backward beyond the condylar process. Capsular projection of the lower incisor forming a conspicuous expansion located generally just behind the base of the coronoid process. Sigmoid notch comparatively shallow and lunar notch not well excavated (Fig. 8I). Upper and lower rami of the masseteric crest converging anteriorly, the lower ramus more developed than the upper one, extended approximately to the same level of the anterior border of the m1 (or slightly behind), and ending at level of the mental foramen (or just above) (Fig. 8J).
Upper incisors opisthodont (angular index lower than 90 • sensu Thomas, 1919) and faced with orange enamel (Fig. 8B). Molars bunodont and pentalophodont (Fig. 9). M1 with procingulum without anteromedian flexus but with a depression on the base of its anterior face and with an enamel island at the level of the anteroloph; anteroloph, parastyle, and mesoloph well-developed, and posteroloph small (only visible in newly erupted molars) (Fig. 9A). M2 with a reduced procingulum but with mesoloph and posteroloph visible, and with an enamel island between protocone and paracone (just ahead of the median mure). M3 comparatively small, with the same general pattern (although with vestigial structures) as the M2 (Fig. 9A). Procingulum of the m1 without anteromedian flexid but with an enamel island; anterolabial cingulid, mesolophid, and posterolophid well developed (Fig. 9B). The m2 with a reduced procingulum, but mesolophid, and posterolophid distinct, and with a penetrating hypoflexid on the labial side. Although with less developed structures, the m3 shows the same general pattern that m2 (Fig. 9B).
For E. macconnelli, considering only the diploid numbers, we can affirm the existence of cryptic species, given that specimens from Peru present 2n=64, FN=64 when compared to specimens from Venezuela that have a totally different diploid number (2n=76, FN=85); additionally, specimens from Brazil with 2n=64, FN=70, show at least one pericentric inversion leading to the different fundamental numbers (Gardner & Patton, 1976;Musser et al., 1998;Patton, Silva & Malcolm, 2000). Our phylogenetic analyses, although not being based on karyotyped specimens, also corroborates the split of E. macconnelli into three distinct lineages: E. macconnelli A from Brazil, E. macconnelli B from Peru, and the specimen CMNH64561 from Suriname. E. emmonsae was also recovered in two major clades. Previous studies indicated that E. emmonsae may also be a species complex (Costa, 2003;Percequillo, 2015), although the diploid number is the same for reported exemplars. A thorough review of these species is clearly necessary, one that would combine multiple character sets, much broader geographic sampling, and, for E. macconnelli, one that would include samples of all known cytotypes.
Regarding the chromosomal evolution of the genus, we associated the diploid and fundamental numbers into the phylogenetic tree. Independently of the diploid number of the outgroup, the split leading to the E. macconnelli clades (2n=76, FN=85;2n=64, FN=64 and 70) and the sister species clade (2n=80, FN=86;2n=76, FN=86;2n=58,60,64, FN=84), can be used to make the following inferences: (i) the difference between the FN=64 and 70 cytotypes in E. macconnelli is due to pericentric inversions involving biarmed chromosomes, given that the diploid number is the same; (ii) comparison between 2n=64, FN=70 and 2n=76, FN=85 forms led us to hypothesize that tandem fusions/fissions have occurred involving acrocentric chromosomes, and/or pericentric inversion and/or Robertsonian rearrangements in only one biarmed chromosome pair, since the former has four and the latter presents five biarmed chromosome pairs in the respective karyotypes; (iii) Euryoryzomys sp. and E. lamia underwent a reduction in the diploid numbers; (iv) the differences between the karyotypes of Euryoryzoms sp. and the four species with 2n=80 is due to Robertsonian rearrangements, as reported by Silva, Percequillo and Yonenaga (2000); and (v) karyotype differentiation of E. lamia involved complex rearrangements, since the three cytotypes showed an elevated number of biarmed chromosomes comparatively to the 2n=80 karyotypes.
Studies applying FISH techniques using species-specific probes (ZOO-FISH) showed a high number of chromosomal rearrangements even in the species with similar diploid numbers in Cerradomys, Hylaeamys and Oligoryzomys, three other oryzomyine genera (Nagamachi et al., 2013;Di-Nizo et al., 2015;Di-Nizo, Ferguson-Smith & Silva, in press). This could also be the case of Euryoryzomys species since herein we hypothesizedbased on chromosome number variation, decrease (or increase) of the diploid numbers and phylogeny-the occurrence of complex and specific chromosome rearrangements. However, it is worth considering that only FISH with specific probes associated with differential chromosome staining will provide refined information on the chromosomal evolution of the genus.
Despite the various taxonomic changes since its original description, most authors today agree on the validity of E. legatus (e.g., Musser & Carleton, 1993;Musser & Carleton, 2005;Weksler, Percequillo & Voss, 2006;Percequillo, 2015;Pardiñas & Ruelas, 2017), and include it as a component of the mammal fauna of Argentina and Bolivia (e.g., Galliari, Pardiñas & Goin, 1996;Anderson, 1997;Salazar-Bravo et al., 2003;Barquez, Díaz & Ojeda, 2006;Teta et al., 2018). Nevertheless, several authors (including some of those who recognize it as a valid entity) emphasized the need for additional studies on the status of this species, especially with respect to E. nitidus. The lack of reciprocal monophyly in previous molecular studies among samples assigned to those nominal forms (in a context of a parapatric distribution) was the main source of uncertainty of its status.
Viewed in an integrated way, our morphometric analyses clearly separated E. legatus from other species of the genus. Univariate analysis shows several skull measurements that confidently distinguished this species from E. nitidus, E. macconnelli, and E. russatus, for which there are more than 13 metric characters that are significantly different. Even E. lamia, the most similar species according to the univariate analysis, shows seven measurements that significantly separated both species. The multivariate analyses also support the morphometric differentiation of E. legatus, not only in size (mainly from E. macconnelli and E. emmonsae), but also in shape of the skull (mainly from E. macconnelli and E. lamia, but also from E. emmonsae and E. nitidus).
The comparison between E. legatus and Bolivian populations assigned to E. nitidus deserves special attention, because of their close phylogenetic relationships. Musser et al. (1998), mostly based on specimens coming from Bolivia, differentiated E. legatus and E. nitidus on morphometric grounds. Results of our analyses, including samples of E. legatus from NW Argentina and southern Bolivia, confirm the distinctiveness between these two species (see also Figs. S5 and S6, and Tables S5-S8). Specimens of E. legatus were, on average, larger than E. nitidus for several measurements (Table 1). Total length (TL), tail length (T), hind foot length (HF), ear length (E), condyloincisive length (CIL), palatal bridge (PB), molar toothrow length (MTRL), bullar length less tube (BLLT), alveolar width (AW1), zygomatic breadth (ZB), zygomatic plate (ZP), braincase breadth (BB), interorbital constriction (IOC), mid rostral width (RW2), rostral length (RL), orbital length (OL), and occipital condyle width (OCW) are significantly larger in E. legatus. In contrast, E. nitidus appears significantly larger than E. legatus for the incisive foramina length (IFL). As revealed by the ''size free'' PCA and DFA, populations of both species also could be separated by the shape of the skull, because only 5% of the specimens of E. legatus were misclassified as E. nitidus and just 1.8% of the specimens of E. nitidus were misclassified as E. legatus. Finally, the chromatic differences in pelage corroborate the distinction of the two taxa.
In addition to the morphological evidence, molecular data also corroborate the separation between E. legatus and E. nitidus. Our phylogenetic results recover E. nitidus as paraphyletic, with genetic distance of 4.5% between the clades E. nitidus A and E. nitidus B, suggesting that two taxonomic entities can actually be considered: one from central Brazil (E. nitidus clade A) and another from western Amazon (E. nitidus clade B). The phylogenetic results point to the segregation of E. legatus from both E. nitidus A and E. nitidus B, and the genetic distances between E. legatus and these two clades are similarly high (E. nitidus A -5.1% and E. nitidus B -4.2%). Nevertheless, despite having expanded the number of specimens for each species, we did not have any E. legatus sample from the sympatric area in Bolivia to be compared. Additionally, our analyses lacked sequences from nuclear genes, because we had access only to cyt-b sequences for key samples (E. nitidus from Bolivia).
Peculiarly, the individual E. nitidus MSB70697 from southern Bolivia was recovered as sister to E. legatus. The collecting locality of MSB70697 (Bolivia: Santa Cruz de la Sierra: 1 km NE Estancia Cuevas; Anderson, 1997) is reported as exhibiting sympatry between E. legatus and E. nitidus (Musser et al., 1998). Although the specimen MSB70697 was identified by Musser et al. (1998) as E. nitidus, it was not included in their morphometric analyses (Musser et al., 1998: 212-213) from which the identification was inferred. We also noticed that this record came from the same type of habitat (montane Yungas forest) occupied by E. legatus (most additional records referred to E. nitidus in Bolivia are from a very different environment, the Chiquitano Forest, pertaining to the Amazonian Domain -see Anderson (1997) and Musser et al. (1998)). Therefore four hypotheses can be postulated for the recovered phylogenetic pattern: (i) MSB70697 is in fact an E. legatus, since this sample had higher similarity to E. legatus (genetic distance of 1.7% to 1.9%) than to E. nitidus (3.7% to 5.3%); (ii) each clade of E. nitidus is a different taxon (E. nitidus A, E. nitidus B, and the MSB70697), rendering E. legatus and each clade within the E. nitidus species complex as monophyletic; (iii) E. nitidus is indeed paraphyletic relative to E. legatus, which can be explained by a recent speciation event with incomplete lineage sorting; and (iv) E. legatus and E. nitidus are indeed the same taxon and thus the former is a junior synonym of the latter. If we integrate all the data obtained herein (morphological analysis, phylogeny, and genetic distance), we can consider one of the first three hypotheses and reject the last one. In this context, at this moment, we suggest maintaining the specific status of E. legatus.
It is clear that further studies investigating these hypotheses must be carried out. Looking at the problem in a broader perspective, these research lines altogether lead us to reinforce the need for a taxonomic revision of Euryoryzomys species based on integrative taxonomy, preferably including broader sampling and particularly sympatric areas, such as the region of Bolivia.

CONCLUSIONS
This is the first study that describes the diploid number of E. legatus and integrates cytogenetics, morphology, and molecular phylogeny to infer the taxonomy and the evolutionary history of Euryoryzomys. Although E. legatus presents 2n=80, FN = 86, the same described for E. emmonsae, E. nitidus, and E. russatus, this karyotype is different from those of E. lamia, E. macconnelli, and Euryoryzomys sp. In addition, E. emmonsae and E. russatus show a disjointed distribution to E. legatus. The species has a close phylogenetic relationship to E. nitidus. We consider E. legatus as a valid species, due to the integration of phylogenetic information, genetic distances, and morphological data. Additionally,